This is kind of long and slightly off topic, but deals with this whole digital quality issue that keeps coming up in audio circles.
If digital transport media and mediums (S/PDIF, AES, DAT, CDs, etc...) are so unreliable (in a computer sense) that you can hear audible differences, how could personal computers and the Internet manage to transfer literally trillions of bytes per second and not grind to a halt? I mean, compared to computer software, financial data, compressed data and encrypted data, audio information is very forgiving.
For example, if CD playback could have enough errors to make an audible difference, does that mean every time I install Microsoft Office, my software will have random defects? Worse yet, since the software is compressed, even a single bit error would clobber large chunks of data because compression relies on past data to contruct current and future data!
After several sleepless nights, I've managed to put my mind at ease. It all comes down to feedback (flow control and error detection / correction). Both of which have been somewhat lacking in consumer audio electronics.
Flow control is the ability to vary the speed with which you receive data. In typical computer fashion, the speeds are typically full tilt and full stop. Using flow control, you can get data faster than you need. But when you can't take any more, you can ask the sender to stop while you digest what you have.
Error correction / detection is pretty much self explanatory. The key points are being able to find an error and deal with it. Either by correcting it yourself or by ask for it again.
Toslink is a good example of something that lacks both. The sender is forced to send data at exactly the same rate as it is needed. If there are any significant errors or timing issues, there is nothing the receiver can do. Even with advanced buffering systems, if the data arrives late more than it does early, the buffer will eventually deplete with no chance of ever filling up. If the data arrives more earlier than late, the buffer will eventually overflow and lose data. However, on average, the buffer can help smooth out small timing flaws. But with errors, there is no recourse except to either play the damaged data or drop it.
With CD-ROMs things are *potentially* much better. They have flow control, error detection and error correction. They typically spin faster than 1x, so you can quickly fill your buffers and then stop reading. In other words, you can reclock the data to eliminate timing problems. This also introduces lots of idle time for the CD drive which can be spent rereading data to detect and recover from bad data (a la EAC).
I use the word potentially because for the sake of cost and political reasons, most consumer grade CD players and other audio equipment lack some or all of the features required for "perfect" CD playback. If any of you want to make a killing on the audiophile CD player market, simply package a CD-ROM drive, EAC and a good DAC into a rackmount chassis and you'd be set.
Audio&Me: If by transport, you mean a S/PDIF like wired transport, I can give you an audiophile quality, errorless, jitterless transport for under $20. It's called Ethernet, you can pick up a 10 Mbit card for dirt cheap and a 100 Mbit one for about the same price now. They have flow control, error detection and enough bandwidth for all of your stereo audio needs. Why, you might ask we don't use Ethernet instead of S/PDIF? Because the RIAA would have fits since anybody with a computer could rip CDs right off their CD players. Oh, wait a minute, there's already consumer level rip and burn machines on the market already
And why does consumer grade "errorless" transports cost upwards of $5000? Because it's extremely expensive to compansate for the flaws in S/PDIF instead of just doing it right the first time.
If by transport, you meant CDs, look no futher than your home PC.